Arxiv:2101.05736V1 [Cond-Mat.Mtrl-Sci] 14 Jan 2021

Metastable piezoelectric group IV monochalcogenide monolayers with a buckled honeycomb structure

Shiva P. Poudel1, ∗ and Salvador Barraza-Lopez1, 2, † 1Department of Physics, University of Arkansas, Fayetteville, AR 72701, USA 2Institute for Nanoscience and Engineering, University of Arkansas, Fayetteville, Arkansas 72701, USA (Dated: January 15, 2021) Multiple two-dimensional materials are being na¨ıvely termed stable on the grounds of displaying phonon dispersions with no negative frequencies, and of not collapsing on molecular dynamics calculations at fixed volume. But, if these phases do not possess the smallest possible structural energy, how does one understand and establish their actual meta-stability? To answer this question, twelve two-dimensional group-IV monochalcogenide monolayers (SiS, SiSe, SiTe, GeS, GeSe, GeTe, SnS, SnSe, SnTe, PbS, PbSe, and PbTe) with a buckled honeycomb atomistic structure–belonging to symmetry group P3m1–and an out-of-plane intrinsic electric polarization are shown to be metastable by three independendent methods. First, we uncover a coordination-preserving structural transformation from the low-buckled honeycomb structure onto the lower-energy Pnm21 (or Pmmn for PbS, PbSe, and PbTe) phase to estimate energy barriers EB that must be overcome during such structural transformation. Using the curvature of the local minima and EB as inputs to Kramers escape formula, large escape times are found, implying the structural metastability of the buckled honeycomb phase (nevertheless, and with the exception of PbS and PbSe, these phases display escape times ranging from 700 years to multiple times the age of the universe, and can be considered “stable” for practical purposes only in that relative sense). The second demonstration is provided by phonon dispersion relations that include the effect of long-range Coulomb forces and display no negative vibrational modes. The third and final demonstration of structural metastability is furnished by room-temperature ab initio molecular dynamics for selected compounds. The magnitude of the electronic band gap evolves with chemical composition. Different from other binary two-dimensional compounds such as transition metal dichalcogenide monolayers and hexagonal boron nitride monolayers which only develop an in-plane piezoelectric response, the twelve group-IV monochalcogenide monolayers with a buckled honeycomb structure also display out-of-plane piezoelectric properties.

I. INTRODUCTION understand metastability, which should be addressed as new 2D materials are laid out. Stereochemistry studies the multiple possible atomistic Indeed, when introducing blue phosphorene (a low- arrangements of chemical compounds. As an incipient buckled honeycomb two-dimensional form of phospho- example of stereochemical behavior, graphite is the most rus), the following is said concerning structural stability: stable carbon allotrope under standard temperature and “blue and black phosphorus are equally stable;” “the rel- pressure conditions. Nevertheless—and despite it not be- ative energy with respect to the black phosphorus struc- ing a structural ground state—diamond is a metastable ture illustrate ... that the blue phosphorus structure is carbon allotrope at room temperature and atmospheric equally stable.” “We find blue phosphorus to be nearly as pressure that can last for over multiple centuries. stable as black phosphorus, the most stable phosphorus References1 and2 and others list a vast amount of yet- allotrope.”16 Germanene and stanene have a nine-fold co- to-be-experimentally synthesized 2D compounds. Along ordination in their most stable two-dimensional phases17. these lines, theoretical predictions of two-dimensional Low-buckled phases, on the other hand, are three-fold 3–7 group-IV monochalcogenides are slowly but surely coordinated. Nevertheless, it has been said that “ger- 8–10 finding experimental confirmation while new and manium can have stable, two-dimensional, low-buckled, remarkable structures—such as one-dimensional, chiral honeycomb structures,” that “the stability of low-buckled 11 GeS —are being experimentally found. Besides the structures of ... Ge are further tested by extensive ab ini- well-known two-dimensional phase with a Pnm21 group tio finite temperature molecular dynamics calculations,” symmetry,4,12,13 group-IV monochalcogenides have also arXiv:2101.05736v1 [cond-mat.mtrl-sci] 14 Jan 2021 and that “the present analysis together with calculated been predicted to form a 2D phase with Pma2 symme- phonon dispersion curves provides a stringent test for the 14 try (SiS), a so-called A−MX phase (whose symme- stability of LB honeycomb structure of ... Ge.”18 Simi- 5 try group was not identified), and a buckled honeycomb larly, one reads that “a low-buckled configuration is found 5–7,15 phase. to be more stable for stanene.”19 In what follows, we focus We make a case for the lack of discussions of structural on diamond and on the low-buckled phases of group IV metastability, by revising the language employed when monochalcogenide monolayers with a buckled honeycomb introducing three extremely popular two-dimensional structure, and demonstrate processes to understand the materials with a low-buckled honeycomb structure. Af- relative stability of phases located at local minima in the terwards, we facilitate processes that can be employed to elastic energy landscape. We then provide a study of 2 the electronic, elastic, and piezoelectric properties of the the path shows tall energy barriers, and escape times are latter compounds. obtained using Kramers escape formula.23 (ii) Phonon GeS, GeSe, SnS, and SnSe monolayers with a buck- dispersion calculations—in which the long-range effec- led honeycomb structure were studied in Ref.5. In that tive charge contribution to the dynamical matrix24 is work, ab initio molecular dynamics (AIMD) within the included—displaying no negative frequencies. (iii) AIMD NVT ensemble (an ensemble with a constant number of calculations within the NPT ensemble at room temper- atoms N, constant volume V , and constant tempera- ature for two such compounds that show small varia- ture T ) were performed for up to 10 ps at room tem- tions in interatomic distances from the reference, zero- perature. Vibrational frequencies at the Γ−point were temperature structure. Chemical composition is shown also calculated, to find no vibrational modes with imagi- to be an additional handle to engineer the magnitude nary frequencies. Binding energies were reported, as well of the electronic band gap. This work ends with a re- as electronic structures that show these materials to be vision of the piezoelectric properties of these materials semiconductors with indirect electronic band gaps. The in Sec.VII, including their out-of-plane piezoelectric re- out-of-plane intrinsic electric polarization, phonon dis- sponse. Conclusions are provided in Sec.VIII. persions, and AIMD calculations using the NVT ensemble were reported in Ref.6 for GeS and GeSe. Duerloo and coworkers calculated the piezoelectric II. COMPUTATIONAL METHODS properties of hexagonal boron nitride (hBN) monolayers and of transition metal dichalcogenide (2H-MoS2, 2H- Ab initio calculations were performed with the VASP MoSe2, 2H-MoTe2, 2H-WS2, 2H-WSe2, and 2H-WTe2) code25,26 on freestanding monochalcogenide monolayers monolayers,20 and piezoelectricity has been experimen- with a buckled honeycomb structure. We employed 21,22 tally confirmed for MoS2 and hBN monolayers since. a 18×18×1 k−point grid and a cutoff energy of 500 Unlike the group-IV monochalcogenide monolayers with eV. The energy and force convergence criteria were set a buckled honeycomb structure studied here, these 2D to 10−6 eV and 10−3 eV/A,˚ respectively. We em- materials are mirror-symmetric with respect to the plane ploy exchange-correlation functionals that include self- defined by the hBN monolayer, or by the plane defined consistent van der Waals corrections27 with the optPBE- by Mo or Te atoms in the 2H-dichalcogenide monolayers, vdW functional.28–30 Dipole corrections were employed and hence do not develop an out-of-plane piezoelectric unless explicitly indicated. response. Phonon dispersion calculations were performed with Using Duerloo and coworkers’ procedure,20 the in- PHONOPY 31 and VASP on a 7×7×1 supercell. Finite plane piezoelectric coefficients were reported for GeS, atomic displacements were set at 0.005 A.˚ In these su- GeSe, SnS, and SnSe monolayers with a buckled hon- percell calculations, the k−point grid was set to 5×5×1 eycomb structure.5 Nevertheless, inversion symmetry is and the cutoff energy remained at 500 eV. The energy broken in this 2D phase, and we will report the resulting convergence remained at 10−6 eV as well. Importantly, out-of-plane intrinsic electric polarization and the out- the effect of Born charges was included in the phonon of-plane piezoelectric response that is missing in Ref.5. dispersion calculations in order to properly describe their The existence of (i) an out-of-plane intrinsic polariza- long-wavelength behavior.24 The magnitude of the polar- tion and of (ii) a tunable electronic bandgap by mate- ization vector P was obtained using the modern theory rial thickness6 or by in-plane strain makes honeycomb of polarization.32 buckled group-IV monochalcogenide monolayers7 rele- Additionally, we performed room-temperature AIMD vant materials for water-splitting applications.6,7,15 As it calculations33 with the SIESTA code34 without dipole will be shown here, further band gap tunability may be corrections for two representative SiS and PbS mono- achieved via chemical composition when the twelve pos- layers for 15 ps on a rectangular supercell containing sible compounds in this family are considered (only five 336 atoms, and AIMD calculations for a PbS mono- such compounds—GeS, GeSe, SnS, SnSe, and GeTe— layer at 2,000 K for 2 ps. This computer code employs have been studied thus far). localized numeric atomic orbitals (NAOs)35 and norm- The manuscript is structured as follows: Computa- conserving Troullier-Martins pseudopotentials36 tuned tional methods are disclosed in Sec.II. Then, the con- in-house37 with van der Waals corrections of the Berland- 38 cepts of the energy barrier EB separating two structural Per Hyldgaard (BH) type as implemented by Román- phases, and of Kramers escape time are exemplified in Pérezand Soler.39 Not needing to fill vacuum with plane a diamond-to-graphite transformation in Sec.III. The waves, the use of NAOs makes AIMD calculations more group symmetries of group-IV monochalcogenide mono- economic; hence this choice of code for these calculations. layers with a buckled honeycomb structure are deter- The AIMD calculations were performed using the NPT mined in Sec.IV, and their structural metastability at ensemble (constant number of electrons N, and at a tar- room temperature is established in Sec.V using three get ambient pressure P at selected temperatures T ) using complementary approaches, including: (i) The determi- methods described elsewhere.40–42 nation of an energy path joining the high-energy P3m1 Periodic images along the direction perpendicular to phase to the low-energy Pnm21 (or Pmmn) structure; the 2D material were separated by a distance a3 = 20 A˚ 3

q in all calculations involving 2D materials. k frequency ω = m at the metastable local minimum centered at point B, to the barrier height EB, and to T , through the following relation:23,46 III. STRUCTURAL METASTABILITY OF 2π EB DIAMOND τ(T ) = λ exp , (1) ω T

We begin by reminding the reader that atomistic con- with EB given in K, as it is the case in Fig.1(a), and figurations having higher energy than ground-state struc- λ being a dimensionless prefactor. We will consider the tures are metastable. A classical example is presented predominant contribution of the exponential to the es- by diamond and graphite, well-known carbon allotropes cape time, and will assume λ to be of the order of unity. available under ordinary temperature and pressure con- The parameter k is obtained by fitting the energy around the local minimum (point B) E = k (c − c )2, where ditions. 2 D ˚ Figure1(a) describes a structural transformation cD = 6.21 A is the magnitude of c for the diamond su- whereby diamond turns into ABC graphite. The left- percell, and m is the mass of the 6-atom supercell. The frequency ω turns out to be 43.4 THz, and the escape most structural diagram atop Fig.1(a) contains side and 9 top views of diamond. There, the z−axis is oriented time is estimated to be 10 years at room temperature. Similar estimates based on the WKB approximation47 along the (111) crystallographic direction on a cell con- 11 taining six atoms; the cell is characterized by lattice yield an even larger τ = 10 years. Given that the age of the universe is roughly 1010 years, such astronomi- parameters a√and c, and atomic√ positions are√ given by (0, 0, 0), (a/ 3, 0, ∆z), (a/ 3, 0, c/3), (2a/ 3, 0, c/3 + cal escape times provide additional rationale for the ap- √ parent stability of diamond, which is additionally con- ∆z), (2a/ 3, 0, 2c/3), and (0, 0, 2c/3 + ∆z). The point firmed by the lack of negative/imaginary frequencies in is that the structure can be fully characterized by three Fig.1(b). Fig.1(c) shows phonon dispersions for ground- independent variables (a, c, and ∆z). Being nonpolar state Bernal graphite which, expectedly, display no neg- compounds, no dipole corrections are necessary at this ative/imaginary frequencies either. stage. One may employ the nudged elastic band method43,44 to estimate the energy barrier in between diamond and IV. STRUCTURE AND CRYSTAL graphite. Nevertheless, such method may overestimate SYMMETRIES OF GROUP-IV the energy barrier (see, e.g., Ref. 45). The alternative MONOCHALCOGENIDE MONOLAYERS WITH approach followed here is a full sampling of the (a,c,∆z) A BUCKLED HONEYCOMB STRUCTURE space over a sufficiently wide parameter range: being explicit, a was sampled from 2.38 to 2.53 A˚ in fifteen steps, c Fig.2(a) depicts a cis-to-trans reconstruction similar ran from 6.0 to 11 A˚ in fifty six steps, and ∆z varied from to the one outlined in Ref. 16 whereby blue phosphorene 0.0 to 0.6 A˚ in fifteen steps, for a total of 12,375 individ- (two-dimensional phosphorus with a buckled honeycomb ual calculations. Any energy value shown in Fig.1(a) is structure) was argued for. This transformation may be acquired on the optimal structure for a given choice of c, relevant to create two-dimensional buckled honeycomb and Fig.1(a) depicts the lowest possible energy barriers phases out of layered group-IV monochalcogenides (SiS, in between diamond and graphite. The additional struc- SiSe, GeS, GeSe, SnS, and SnSe). Two-dimensional tural plots (B to G) in Fig.1(a) illustrate the structural buckled honeycomb compounds could be grown along the evolution of the six-atom cell into ABC graphite (G) as c (111) directions on materials with rhombic or cubic bulk increases; capital letters match structures seen on top to structures (SiTe, GeTe, SnTe, PbS, PbSe, and PbTe)48 the points shown in red along the energy versus c path in on suitable commensurate substrates. Growth techniques the Figure. Na¨ıvely, the lower energy ∆E = −8, 150 K advance at a fast pace, and two-dimensional films can of graphite with respect to diamond may imply that di- be created by growth and subsequent wet chemistry.49 amond is not stable. Nevertheless, a large EB of 19,881 Fig.2(c) shows another possible structure which may K is the key energetic variable guaranteeing diamond’s compete with the buckled honeycomb one: a 2H dichalco- structural metastability. genide, which has been reported experimentally50–52 and The inset in Fig.1(a) depicts a horizontal line that rep- theoretically53 already. Similarly, Fig.2(d) illustrates the resents the—comparatively tiny—energy difference be- creation of a 1T dichalcogenide by the subsequent growth tween ABC and Bernal graphite (the energy of Bernal of a chalcogen layer on the film grown along the (111) graphite—whose unit cell contains four atoms—has been direction of monochalcogenides with a non-layered bulk multiplied by 3/2 for a direct comparison). A stan- structure. We acknowledge that the buckled honeycomb dard construct in the theory of glasses—i.e., materials structures have not been observed in experiment, and the that display multiple local minima—is Kramers escape challenges that their realization entails. formula,23 which permits estimating escape times τ from The group symmetry of buckled honeycomb group- a metastable minimum onto the ground state structure IV monochalcogenides is derived here following a pro- qualitatively; τ is proportional to the classical oscillation cess that starts with graphene’s symmetry operations 4

FIG. 1. Computational demonstration of the metastability of diamond. (a) When the z−axis is oriented along the (111) crystallographic direction, diamond can be turned into ABC graphite by following the lowest-energy structural path depicted by the structures shown above. (The structural energy of Bernal graphite is seen as a horizontal straight line at the inset.) Despite of having an energy 8,150 K higher than ABC graphite, it takes an astronomical time of 109 years for diamond to tunnel along the horizontal dashed line through the energy barrier EB . (b) The large barrier height guarantees the existence of phonon dispersions with no imaginary frequencies for diamond. (c) Phonon dispersion for the ground-state carbon allotrope, Bernal graphite.

(symmetry group p6mm). According to the Interna- symmetry operations are x, y;y, ¯ x − y;x ¯ + y, x¯;y, ¯ x¯; tional Tables for Crystallography, and as shown in Fig.3, x¯ + y, y; and x, x − y; which correspond to symmetry the standard choice for lattice vectors is one in which group p3m1.54 Coordinates x = 2/3, and y = 1/3 √ A A 1 3 54 (x = 1/3, and y = 2/3) give atomic positions of the A a1 = a − 2 , − 2 , 0 and a2 = a (1, 0, 0). In stan- B B dard Crystallography notation, atomic coordinates are (B) sublattice. expressed in direct format (cartesian positions of atoms The main structural diﬀerence between a hBN mono- are given by xa1 + ya2 in 2D, or xa1 + ya2 + za3 in layer and buckled honeycomb group-IV monochalco- 3D). The twelve symmetry operators for graphene (space genide monolayers is the out-of-plane buckling in the lat- group 17) have the following coordinates: x, y (symmetry ter, which requires supplementing the 2D coordinates of operation 1);y, ¯ x − y (2);x ¯ + y, x¯ (3);x, ¯ y¯ (4); y, x¯ + y the hBN monolayer with a relative height. We write (5); x − y, x (6);y, ¯ x¯ (7);x ¯ + y, y (8); x, x − y (9); y, x xA = 2/3, yA = 1/3, zA = zIV /a3, and xB = 1/3, 54 (10); x − y, y¯ (11); andx, ¯ x¯ + y (12). (Here,x ¯ = −x yB = 2/3, zB = zVI /a3 with zIV and zVI atomic heights 54 andy ¯ = −y. ) Taking x = 2/3 and y = 1/3, symme- in A˚ such that ∆z = a3(zB − zA). The allowed symmetry operations (1), (2), (3), (7), (8) and (9) leave atoms try operations turn out to be: x, y, z (1);y, ¯ x − y, z (2); within the same sublattice (say, A), while atoms in the x¯+y, x,¯ z (3);y, ¯ x,¯ z (4);x ¯+y, y, z (5); and x, x−y, z (6), 1 54 B sublattice are reached through symmetry operations corresponding to symmetry group 156 (P3m1, or C3v). (4), (5), (6), (10), (11) and (12) from an atom originally Group P3m1 has a three-fold rotational symmetry belonging to the A−sublattice. (shown by green axial lines in Fig.3, top view), a mir- Turning into a hBN monolayer, the six symmetry op- ror symmetry shown by a black dotted line, and it lacks erations above [(4), (5), (6), (10), (11), and (12)] taking an out-of-plane inversion symmetry, as clearly observed a boron atom (A−sublattice) and landing it into a nitro- in both side views on that Figure. Symmetry operations gen atom (B−sublattice) are forbidden. The six allowed (1) through (6) were explicitly applied to a SiS monolayer 5

FIG. 2. Conceptual rendering of group-IV monochalcogenide monolayers with a buckled honeycomb structure: (a) Lay- ered group-IV monochalcogenides (SiS, SiSe, GeS, GeSe, SnS, and SnSe) on the unit cell with Pnm2 symmetry (horizontal 1 FIG. 3. Structure and symmetry operations of buckled structure) and on a buckled honeycomb structure reminis- honeycomb group-IV monochalcogenide monolayers. The top cent of blue phosphorene. (b) Rhombic and cubic group-IV view lists lattice vectors, the unit cell area (in orange), atomic monochalcogenides (SiTe, GeTe, SnTe, PbS, PbSe, and PbTe) positions, symmetry operations, a three-fold rotation axis (in grown along the (111) direction can also give rise to buckled dark green), and a mirror plane (dotted black line). Atoms honeycomb structures if grown on an appropriate substrate. belonging to the A (B) sublattice are highlighted with open (c and d) Additional possible structures could be created, too. black (full red) circles. A rectangular cell containing four The unit cell area is highlighted within orange diamonds in atoms is drawn in light green at the upper left. ∆z is de- subplots (b) and (d). h picted in the side views, which lack inversion symmetry. in Fig.3, top view, and TableI lists the lattice constant V. STRUCTURAL METASTABILITY ah and a (signed) buckling height ∆zh of the optimal buckled honeycomb (h) structures. The lack of inversion The methods illustrated in Fig.1 plus AIMD will now symmetry confers these materials with an intrinsic out- be applied to establish the structural metastability of the of-plane electric intrinsic polarization P 6 (also listed h,3 buckled phase of two-dimensional group-IV monochalco- in TableI), and an out-of-plane piezoelectric coefficient genides. that is missing in Ref.5. Lattice constants are compara- ble to these reported previously, with differences arising from the different choice of exchange-correlation functional (vdW here, and PBE55 in previous reports; see A. Local minimum and Kramers escape times TableI). The magnitude of the lattice constant on a planar structure, ap is also listed in TableI for reasons that In direct analogy to the process we followed to deter- will become apparent latter. The average atomic number mine the energy barrier EB underpinning the transforma- ¯ ZIVA+ZVIA Z = 2 [with ZIVA (ZVIA) the atomic number tion from diamond to graphite, we first envision a struc- of the group-IVA (VIA) atom] is listed in subsequent Fig- tural transformation that preserves the three-fold atom- ures and Tables, to emphasize structural trends in this istic coordination, and converts a structure with P3m1 40,56 family of compounds that depend on that variable, symmetry onto the lower-energy Pnm21 one (or Pmmn ¯ such as the increase of ah, |∆zh|, ap, and Ph,3 with Z in for PbS, PbSe, and PbTe for reasons to be explained TableI. later on). This transformation is, in fact, the cis-to-trans When dealing with two-dimensional materials, it has conformal change alluded to in Fig.2(a). become customary to list polarization in units of C/m As it turns out, the transformation traverses through (see, e.g., Refs.5, 20, 41, and 57, among many others) a planar structure (∆z = 0) with p3m1 symmetry, whose or in Debye per unit cell (D/u.c.) as in Ref.6. C/m lattice constant ap is listed in TableI as well. The lack of units are employed here for direct comparison with the inversion symmetry with respect to the XY −plane leads relevant literature. to a net intrinsic electric polarization P3 for ∆z ≤ 0, 6

Y Y Top Side 1 X X

Δz Δz A B C D Z E F G H Side 2 X

E p p 1,200 30,000 SiS SiSe SiTe 800 D 400 F (15) (24) (33) 0

20,000 planar 2.646 2.658 A 10,000 C G honeycomb rectangular h h r r 0 B H 30,000 GeS GeSe p 800 GeTe p 400 400 200 20,000 (24) p (33) (42) 0 0 2.48 2.51 2.54 2.72 2.75 2.78

10,000 h h h r r r 0 30,000

Energy (K) 180 300 p p 30 SnS SnSe 120 SnTe 150 (33) p (44) 60 (51) 15 20,000 0 0 2.586 2.592 2.7305 2.7355 2.948 2.950 h 10,000 h h r r r 0 30,000 p PbS p PbSe p PbTe 20,000 (49) (58) (67) h EB h h 10,000

ΔE square s s 0 -2.0 -1.0 0.0 1.0 2.0 3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 ΔZ (Å)

FIG. 4. Top: Selected atomistic conﬁgurations for a coordination-preserving transformation from a buckled honeycomb structure A onto a unit cell with Pnm21 symmetry, as exempliﬁed on a SiS monolayer. A rectangular cell containing four atoms is drawn in light green. Bottom: Energy versus ∆z for twelve two-dimensional group-IV monochalcogenide monolayers. ∆E, EB , and the locations along the path of the local minimum buckled honeycomb (h, structure B), planar (p, structure E), and ground state rectangular (r, structure H) unit cell with Pnm21 symmetry are highlighted. Insets show the energy cost to turn the rectangular unit cell into a square unit cell with Pmmn symmetry. PbS, PbSe, and PbTe have a ground state structure with Pmmn symmetry hosting a square (s) unit cell.

0 0 whose magnitude is depicted in Fig.5(b) ( X, Y , and Z tudes of a1 and a2 are provided as Supplemental Material axes are illustrated in Fig.4, structure E). The in-plane in Ref. 56); (iii) a glide-reflection plane M¯ XY : a reflec- three-fold and mirror symmetries of the P3m1 symme- tion by the XY plane followed by R; and (iv) a reflec- 13 try group are preserved in this structural transformation, tion about the xz plane (MXZ ) . The glide-reflection rendering P1 and P2 equal to zero. (When piezolelectric symmetry quenches the out-of-plane intrinsic polariza- properties are considered later on, the in-plane strain will tion (P3 = 0) for ∆z > 0, while the mirror symmetry break the in-plane three-fold symmetry, hence inducing along the XZ plane renders P2 = 0 for all values of a non-zero P1 and P2.) ∆z in Fig.5. Structures F to H show the process in which nearby atoms buckle in opposite directions (this The low-energy Pnm2 phase has the following sym- 1 is why the two-atom honeycomb structure turns into a metries: (i) the identity E; (ii) C¯ : a two-fold rota- 2X four-atom unit cell), explaining in graphical terms why tion around the X−axis (C ), followed by a translation 2X P for ∆z ≥ 0. We used the pair of atoms highlighted R = (a0 + a0 )/2 where a0 = (a0 , 0, 0) and a0 = (0, a0 , 0) 3 1 2 1 1 2 2 in solid red and open black colors to define ∆z; this pair are the lattice vectors of the rectangular cell (the magni- 7

TABLE I. Optimized lattice constant ah, signed buckling TABLE II. Energy difference ∆E, energy barrier EB , and ω height ∆zh, and out-of-plane intrinsic polarization Ph,3 for for two-dimensional group-IV monochalcogenide compounds. group-IV monochalcogenide monolayers with a buckled hon- Compound Z¯ ∆E (K) EB (K) ω (THz) eycomb structure. The optimized lattice constant for a planar P3m1 to Pnm21 SiS 15 −858 32,927 9.1 structure (ap) is listed as well. Additional data from the lit- P3m1 to Pnm21 SiSe 24 −115 27,749 6.6 erature was added for a direct comparison. P3m1 to Pnm21 SiTe 33 −1,271 25,630 5.1 Compound Z¯ ah ∆zh ap Ph,3 Ph,3 P3m1 to Pnm21 GeS 24 −2,511 24,230 6.4 (A)˚ (A)˚ (A)˚ (pC/m) (D/u.c.) P3m1 to Pnm21 GeSe 33 −1,794 22,282 5.2 SiS 15 3.342 −1.328 4.242 −0.7 −0.02 P3m1 to Pnm21 GeTe 42 −2,344 21,849 4.2 SiSe 24 3.551 −1.425 4.451 −1.2 −0.04 P3m1 to Pnm21 SnS 33 −5,358 20,461 4.7 SiTe 33 3.858 −1.543 4.758 −2.0 −0.08 P3m1 to Pnm21 SnSe 42 −3,979 19,731 4.1 GeS 24 3.524 −1.370 4.424 9.5 0.31 P3m1 to Pnm21 SnTe 51 −3,512 20,617 3.5 3.495∗ −1.363∗ — —— P3m1 to Pmmn PbS 49 −11,016 12,393 3.4 3.489† —— — 0.21† P3m1 to Pmmn PbSe 58 −9,023 13,194 3.0 3.495§ —— —— P3m1 to Pmmn PbTe 67 −7,625 15,185 2.7 GeSe 33 3.697 −1.466 4.597 7.5 0.27 3.674∗ −1.451∗ — —— 3.663† —— — 0.16† TABLE III. Estimated escape times τ for group-IV 3.676§ —— —— monochalcogenides with a (P3m1) buckled honeycomb struc- GeTe 42 3.980 −1.579 4.915 4.6 0.19 ture. (The age of the universe is of the order of 1010 years.) ‡ ‡ 3.96 −1.57 — —— Compound Z¯ τ (years) Compound Z¯ τ (years) SnS 33 3.769 −1.473 4.719 13.3 0.49 SiS 15 1.0×1028 SnS 33 1.8×1010 ∗ ∗ 3.753 −1.465 — —— SiSe 24 4.5×1020 SnSe 42 1.8×109 SnSe 42 3.922 −1.575 4.922 9.3 0.37 SiTe 33 5.0×1017 SnTe 51 4.0×1010 ∗ 3.910 −1.564 — —— GeS 24 3.7×1015 PbS 49 0.05 SnTe 51 4.193 −1.710 5.243 7.3 0.33 GeSe 33 7.0×1012 PbSe 58 0.80 PbS 49 3.944 −1.468 4.844 24.6 1.00 GeTe 42 2.0×1012 PbTe 67 7.1×102 PbSe 58 4.064 −1.593 5.014 21.5 0.93 PbTe 67 4.323 −1.722 5.373 17.2 0.83 ∗ † : PBE, VASP; Ref.5. : PBE, VASP (DF3 van der Waals second portion, in white color, takes the p3m1 structure corrections for few-layer stacks); Ref.6. ‡: PBE, VASP onto either the Pnm2 structure (for SiS, SiSe, SiTe, GeS, (DF3 van der Waals corrections for few-layer stacks); 1 Ref. 15. §: PBE, VASP; Ref.7. GeSe, GeTe, SnS, SnSe, and SnTe) or the Pmmn structure (for PbS, PbSe, and PbTe). Figure4 demonstrates, without exception, that the energy barrier EB takes its swaps the sign ∆z on structures F to H with respect to maximum value for the planar structure. Values of EB, its value in structures A to D, hence giving the change of ∆E, and ω are listed in TableII. We note that EB is of sign on the ∆z parameter we used to describe the trans- the same order of magnitude than the one listed for the formation. diamond to graphite transformation, and Eqn. (1) yields escape times ranging from a month for PbS, up to multi- A transformation of the Pnm21 unit cell into a four-fold symmetric (square, s) Pmmn structure with a net P = 0 ple ages of the universe for SnTe, GeTe, GeSe, SiTe, SiSe, takes place within the dark-blue section of the energy ver- and SiS monolayers with a buckled honeycomb structure sus ∆z plots for nine compounds (SiS, SiSe, SiTe, GeS, (TableIII). GeSe, GeTe, SnS, SnSe, and SnTe)40,42,56,58 for which the Pmmn structure has a larger energy, as verified by the increase in energy in Fig.4. Zoom-ins were inserted to 1. Evolution of polarization along the structural some subplots to emphasize the increase in energy. Two- transformation dimensional PbS, PbSe, and PbTe display a global minimum with Pmmn symmetry already;40,42,56,58 and ∆z Figure5 depicts the evolution of the intrinsic elec- was just enlarged in these structures to highlight such tric polarization P = (P1,P2,P3) of the SiS monolayer (global) minima. More specific technical details of the along the ∆z path depicted in Fig.4. From left to transformation can be found as Supplemental Material59. right, vertical dashed lines label structures h, p, and r Here, it suffices to say that the energies shown in Fig.4 (B, E, and H in Fig.4, respectively). Structures with a take on their minimum possible magnitudes for fixed val- honeycomb unit cell (∆z ≤ 0) are three-fold symmetric ues of ∆z, following an explicit optimization procedure and hence lack an intrinsic in-plane electric polarization 60 similar to the one employed in Fig.1. (P1 = P2 = 0 in Fig.5). Shaded vertical areas indicate Summarizing, the energy versus ∆z plots are divided structures that yield a metallic electronic structure59 for in three sections. The first one, colored in light blue, which the polarization cannot be computed within the corresponds to the P3m1 to p3m1 (buckled honeycomb standard approach,32 leading to a slight discontinuity for to planar honeycomb) part of the transformation. The P1 for ∆z in between 0 and 0.5 A.˚ 8

400 15 P with a buckled honeycomb structure. For instance, the (a) (b) 2 300 P matrix coupling among the central Si atom and a ﬁrst- 10 1 neighbor S atom (located to its left along the x−axis and 200 a distance |∆zh| above) is given by: 5 P3 (pC/m) 100 3 −2.7497 0.0000 3.6220  , P

2 0

P KSiS(1) =  0.0000 −0.8307 0.0000 

(pC/m) 0 1 1.8594 0.0000 −3.6082 P P -5 2 -100 P3 2 P1 in units of eV/A˚ . The coupling among in-plane and -200 -10 -2.0 -1.0 0.0 1.0 2.0 3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 out-of-plane directions has an obvious origin: a horizon- ∆z (Å) tal relative displacement necessarily modiﬁes the relative height among atoms, and viceversa. In graphene, on the FIG. 5. Evolution of the intrinsic electric polarization P = other hand, a horizontal displacement does not neces- (P1,P2,P3) for a SiS monolayer along the structural transfor- sarily require a change on the relative vertical distances mation depicted in Fig.4. The light and bold blue shading among atoms. The second reason for the discrepancy are consistent with the structural evolution depicted in Fig.4. among the dispersion of the lowest energy phonon band From left to right, the vertical dashed lines correspond to and the one seen in graphene has to do with the low wave- the local minima buckled honeycomb structure (h), the pla- length behavior, which couples to the long-range electro- nar structure with p3m1 symmetry (p), and the rectangular static interaction among ions in these binary materials: structure with Pnm2 symmetry (r). The directions of the 1 as indicated in Sec.II, the eﬀect of Born charges turned polarization vector are displayed as an inset in subplot (b). crucial to describe the long-wavelength vibrational behavior appropriately.24 Lacking negative frequencies, and providing additional credence to the escape time argu- Nevertheless, and as indicated in Ref.6 and observed ment provided in TableIII, the phonon dispersion calcu- in Fig.5(b), an out-of-plane intrinsic polarization P en- 3 lations in Fig.6 represent the second proof of structural sues in binary compounds with P3m1 group symmetry. metastability of group IV monochalcogenide monolayers The magnitude of P at the local buckled honeycomb 3 with a buckled honeycomb structure. energy minimum is labeled P and its magnitude is h,3 It is illustrative to look at the vibrational spectra at listed in TableI for the twelve studied compounds. In the top of the barrier (point p in Fig.4), and Fig.7 order to understand the sign of P , we remark that all 3 serves this purpose. All materials, with the exception compounds with a P3m1 symmetry have their chalco- of SiS, show a single negative mode at the Γ-point, cor- gen atoms below group-IVA atoms. As seen in Fig.5(b), responding with an out-of-plane optical vibration, and the magnitude of P on the P3m1 phase is tunable by 3 thus verifying that point p indeed corresponds with the a change in structure (∆z). This fact is an incipient maximum value of the energy barrier. SiS was known demonstration of out-of-plane piezoelectric behavior, a to have another ground state since earlier versions of discussion omitted in previous work5 that will be pro- this manuscript (Reference 14); that compound displays vided in Sec.VII. Belonging to another symmetry group, an out-of-plane optical vibration, and two degenerate in- structures labeled with ∆z > 0 suppress their out-of- plane vibrations accordingly. plane polarization P3, and may develop a net in-plane 41 polarization P1. [The exception is the Pmmn (s) phase 41,42,56,58 for which P1 = P2 = P3 = 0. ] C. Ab initio molecular dynamics

TableII indicates that SiS has the highest energy B. Phonon dispersion calculations barrier EB, and PbS has the lowest one among the twelve compounds studied. Therefore, if PbS turns to be The phonon dispersions shown in Fig.6 were computed metastable at room temperature in AIMD calculations, at the local energy minima (labeled h) in Fig.4. The one may be able to conclude that the remaining eleven lower band at the Γ−point does not have a quadratic compounds should be metastable, too. Such assertion dispersion, like the one seen in graphene. There are two could be checked by performing AIMD on a second com- reasons for such a discrepancy. Direct inspection into (i) pound, say SiS. the force constant tensor and (ii) into the disposition of It is important to be judicious when choosing the atoms in the supercell from which phonons are computed type of molecular dynamics calculations to be performed: permits observing the force matrices for pairs of atoms structural transformations that require the materials’ lying along the x−axis explicitly—and corresponding to area to change—such as the one illustrated in Fig.4— ﬁrst, third, ﬁfth, and eight nearest neighbors. These may display a forced stability in NVT AIMD calculations matrices are diagonal in graphene (see Equation 9.9 in at a given temperature, because of the constant area con- Ref. 61), but they couple substantially along the x− and straint, that is inconsistent with the structural transfor- z−directions in group-IV monochalcogenide monolayers mation (observe the change in area on the top views in 9

500 K SiS SiSe SiTe M 400 (15) (24) (33) Γ 300

200

100

0 400 GeS GeSe GeTe (24) (33) (42) 300

200

) 100 − 1

SnS SnSe SnTe 300 (33) (44) (51) Frequency (cm

200

100

0 300 PbS PbSe PbTe (49) (58) (67)

200

100

0 Γ K M Γ Γ K M Γ Γ K M Γ

FIG. 6. Phonon dispersion for the twelve group-IV monochalcogenide monolayers with a buckled honeycomb structure at the local energy minima (h) depicted in Fig.4: no negative modes can be seen, and these results constitute the second proof of the structural stability of group-IV monochalcogenide monolayers with a buckled honeycomb structure. The blue dashed curves were obtained with dipole corrections.

Fig.4). In the structural transformations studied by this for the AIMD calculations reported here. team, we have shown that structural instabilities are trig- The distributions of interatomic distances at 300 K for gered at lower temperatures when employing the NPT NPT AIMD calculations on SiS or PbS supercells con- ensemble,58 as the area of these 2D materials is allowed taining 336 atoms, and running for up to 15 ps, are shown to change. This explains our choice of an NPT ensemble in Fig.8. The peaks of these distributions match the 10

400 SiS SiSe SiTe (15) (24) (33) 200

K M -200

-400 400 GeS GeSe GeTe (24) (33) (42) 200

-200 ) - 1

-400 400 SnS SnSe SnTe (33) (44) (51) Frequency (cm

200

-200 400 PbS PbSe PbTe (49) (58) (67)

200

-200 Γ K M Γ Γ K M Γ Γ K M Γ

FIG. 7. Phonon dispersion for the twelve group-IV monochalcogenide monolayers with a planar honeycomb structure at the energy maxima (p) depicted in Fig.4 (no dipole corrections are required given that the structure is planar): With the exception of SiS, all materials show a single negative mode at the Γ-point, corresponding with an out-of-plane optical vibration, thus verifying that point p indeed corresponds with the maximum value of the energy barrier. 11

FIG. 8. Distribution of interatomic distances at 300 K for the (a) SiS monolayer and (b) PbS monolayer with a buckled honeycomb structure. The dashed vertical lines—which match the distribution’s peaks—are interatomic distances at zero temperature. AIMD thus conﬁrms the structural metastabil- FIG. 9. (a) Distribution of interatomic distances at 2,000 ity of these buckled honeycomb phases at room temperature. K for the PbS monolayer with a buckled honeycomb structure. The dashed vertical lines are interatomic distances at zero temperature, and the solid vertical lines correspond to interatomic distances of the planar structure. The sharp peak TABLE IV. Melting temperature TM of bulk group-IV at 2.8 A˚ is a sign of dimerization, that is conﬁrmed by the monochalcogenides according to Ref. 62. (Data for Si-based presence of dimers within red circles in structural snapshots compounds is lacking.) at 1.0 (subplot b), 1.5 (subplot c), and 2.0 ps (subplot d),

Compound Z¯ TM (K) Compound Z¯ TM (K) respectively: PbS turns amorphous at T=2,000 K. GeS 24 938 SnTe 51 1,063 GeSe 33 948 PbS 49 1,383 GeTe 42 997 PbSe 58 1,355 pears to be consistent with a melting temperature TM SnS 33 1,153 PbTe 67 1,197 not larger than 1,200 K (TableIV) for the corresponding SnSe 42 1,153 bulk phase. One then concludes that the energy barrier EB is too tall for all twelve compounds for the elastic transformation depicted in Fig.4 to take place solely by zero-temperature interatomic distances shown by verti- temperature. cal dashed lines. The supercell size and runtimes in our calculations are much larger than these employed in earlier works,5,6 and these AIMD results contribute the third VI. ELECTRONIC BAND STRUCTURES and ﬁnal demonstration of structural metastability of this family of compounds at room temperature. Now that the metastability of these honeycomb phases The next question concerns the feasibility of realizing has been demonstrated using three independent meth- the elastic structural transformation depicted in Fig.4 ods, it is time to turn to the electronic properties of these thermally: after all, the Pnm21 to Pmmn transforma- compounds, which were obtained with the spin-orbit cou- tion can be achieved with temperature.8,40–42,58 Figure pling turned on (in black), and with the HSE06 hybrid 9 shows that the barrier is too high, such that ther- functional63 without spin-orbit coupling turned on (in mal ﬂuctuations in an AIMD calculation at 2,000 K turn cyan) and displayed in Fig. 10. (HSE06 corrections pro- the honeycomb PbS structure onto an amorphous mate- vide bandgaps in closer agreement with experimental es- rial that even undergoes dimerization. This result ap- timates.) With the exception of PbSe and PbTe, these 12

6 SiS SiSe SiTe 4 (15) (24) (33)

-2

-4 6 GeS GeSe GeTe 4 (24) (33) (42)

-2

(eV) -4 F 6

E-E SnS SnSe SnTe 4 (33) (44) (51)

-2

-4 6 PbS PbSe PbTe 4 (49) (58) (67)

-2

-4 Γ K M Γ Γ K M Γ Γ K M Γ

FIG. 10. Electronic band structures with spin-orbit coupling included (in black) for the twelve group-IV monochalcogenide monolayers with a buckled honeycomb structure at the local energy minima (h) depicted in Fig.4. The band structures in cyan were obtained with the HSE06 functional and without spin-orbit coupling. 13

HSE TABLE V. Electronic band gaps Eg and Eg for group-IV TABLE VI. Relaxed-ion, in-plane elastic coefficients C11 and monochalcogenides with a (P3m1) buckled honeycomb struc- C12 (in N/m) for group-IV monochalcogenides with a (P3m1) ture. When available, additional data from the literature was buckled honeycomb structure. When available, additional added for a direct comparison. data from the literature was added for a direct comparison. ¯ HSE Compound Z¯ C C Compound Z Eg (eV) Eg (eV) 11 12 SiS 15 2.41 3.24 SiS 15 49.37 8.36 SiSe 24 2.23 3.03 SiSe 24 40.88 8.59 SiTe 33 1.65 2.47 SiTe 33 36.64 6.62 ∗ ∗ GeS 24 2.59, 2.47∗, 2.49§ 3.40, 3.27†, 3.27§ GeS 24 37.34, 50.63 8.85, 10.78 ∗ ∗ GeSe 33 2.27, 2.27∗, 2.29§ 3.08, 3.01†, 2.99§ GeSe 33 29.73, 49.54 7.29, 10.42 ‡ GeTe 42 29.04 6.33 GeTe 42 1.56, 2.34, 2.35 ∗ ∗ ∗ SnS 33 27.39, 41.59 7.66, 10.52 SnS 33 2.43, 2.31 3.14 ∗ ∗ SnSe 42 2.27, 2.21∗ 2.97 SnSe 42 24.27, 38.99 6.74, 8.56 SnTe 51 1.62 2.44 SnTe 51 21.77 6.72 PbS 49 2.03 2.85 PbS 49 21.33 6.05 PbSe 58 1.68 2.59 PbSe 58 17.04 5.28 PbTe 67 1.10 2.12 PbTe 67 17.77 4.77 ∗: PBE, VASP; Ref.5. ∗: PBE, VASP; Ref.5. †: PBE, HSE06 functional for electronic properties, VASP (DF3 van der Waals corrections ‡ for few-layer stacks); Ref.6. : PBE, HSE06 functional for direction, which yields: electronic properties, VASP (DF3 van der Waals corrections for few-layer stacks); Ref. 15. §: both PBE and HSE06 C ∆E( , 0) = 11 2 . (3) results were reported using VASP; Ref.7. 11 2 11 The quadratic coefficient in Eq. (3) shown in Fig. 11 is thus C /2. P3m1 two-dimensional phases feature indirect bandgaps, 11 C , in turn, is obtained by applying biaxial strain and DFT band gaps within 2.12 and 3.40 eV when HSE06 12 ( , ) with = , which yields: corrections are considered (see TableV): the electronic 11 22 22 11 2 2 2 band gap is a relevant parameter for applications argued ∆E(11, 11) = C11 + C12 = 2∆E(11, 0) + C12 , 6,7 11 11 11 for in previous work, and its magnitude can be fur- (4) ther tuned by the choice of chemical elements in these where the presence of a contribution from uniaxial strain compounds. is made explicit. A quadratic fitting of Eq. (4) yields 2 ∆E(11, 11) = α11 as shown in Fig. 12. Using Eqns. (3) and (4) we get: VII. ELASTIC AND PIEZOELECTRIC 2 2 C1211 = ∆E(11, 11)−2∆E(11, 0) = (α−C11)11. (5) PROPERTIES It is customary practice to divide these coefficients by the equilibrium area of the unit cell, and to write these The in-plane symmetry operations of hexagonal boron constants in units of N/m. The area of the√ four atom nitride (wallpaper, two-dimensional, symmetry group 2 p3m1) apply to the buckled hexagonal compounds, but unit cell at the local minimum is given by 3ah, with the latter lacks out-of-plane inversion symmetry and thus ah given in TableI . The constants obtained from the belong to the (three-dimensional) symmetry group P3m1 quadratic fitting as shown Fig. 11 and Fig. 12 are thus 1.602×10−19C/e √ (Sec.IV). The three-fold symmetry of these compounds multiplied by 2 to get the values listed in 3ah 20 C11−C12 64 renders C22 = C11, and C66 = 2 . (C66 = C1212 TableVI. is the elastic coefficient due to in-plane shear strain Elastic and piezoelectric properties have a strong de- 12, C21 = C12, and Voigt notation has been employed pendence on the choice of exchange-correlation func- throughout.) The magnitudes of an elastic coefficients tional, because forces and dipole moments depend di- C11 and C12 are calculated using a fully relaxed atomic rectly on the charge distribution. Even though it is the configuration( relaxed ion) and listed in TableVI. default choice for many calculations, the PBE functional Elastic constants are calculated following the proce- was never developed to account for materials with a large dure established by Duerloo and coworkers:20 A rectan- amount of vacuum, while ab initio vdW functionals are gular unit cell with four atoms is set, and the following explicitly developed to account for drastic changes in den- expression is employed: sity at interstitials and vacuum regions.27 As indicated before, the elastic coefficients reported in TableVI were C obtained under relaxed-ion conditions; i.e., atomic posi- ∆E( , ) = 11 2 + 2 + C . (2) 11 22 2 11 22 12 11 22 tions were optimized as strain was applied. Materials with a P3m1 group symmetry follow the in- C11 is obtained by applying uniaxial strain along the 11 plane symmetry rules that were developed in Ref. 20, but 14

data 0.003 fit ∆E(ε ,0)=(27.99 eV)ε 2 ∆E(ε ,0)=(29.53 eV)ε 2 2 11 11 11 11 ∆E(ε11,0)=(29.75 eV)ε11 0.002

0.001 SiS SiSe SiTe 0.000

0.003 2 2 2 ∆E(ε11,0)=(25.08 eV)ε11 ∆E(ε11,0)=(22.79 eV)ε11 ∆E(ε11,0)=(24.86 eV)ε11

0.002

0.001 GeS GeSe GeTe 0.000

E (eV) 0.003

∆ 2 2 2 ∆E(ε11,0)=(21.06 eV)ε11 ∆E(ε11,0)=(20.75 eV)ε11 ∆E(ε11,0)=(20.85 eV)ε11 0.002

0.001 SnS SnSe SnTe 0.000

0.003 2 2 2 ∆E(ε11,0)=(17.97 eV)ε11 ∆E(ε11,0)=(15.29 eV)ε11 ∆E(ε11,0)=(17.93 eV)ε11

0.002

0.001 PbS PbSe PbTe 0.000 -0.01 -0.005 0.00 0.005 0.01 -0.01 -0.005 0.00 0.005 0.01 -0.01 -0.005 0.00 0.005 0.01 ε11

FIG. 11. Quadratic ﬁtting of energy versus uniaxial strain 11. The coeﬃcient shown at each plot is 2C11.

out-of-plane piezoelectric properties ought to be added to the out-of-plane piezoelectric responses e31 = e32, which the discussion of these novel phases. The symmetries of were not studied in previous work.5 the P3m1 group render the piezoelectric coefficients e , 21 The e coefficient is negative because the buckling e , e , and e equal to zero because mirror symme- 31 22 16 36 height ∆z turns closer to zero for tensile strain, hence try is preserved upon the application of normal strain, quenching the intrinsic electric polarization [see struc- and because the in-plane and out-of-plane polarization tures A to E in Fig.4, and P in Fig.5(b)]. ∆ P versus are not sensitive to the sign (direction) of the in-plane 3 i strain plots—whose slopes are the non-zero piezoelectric shear strain. The e piezoelectric coefficient is different 11 coefficients e and e listed in TableVII—are shown in from zero, and e = e = −e . Here, the atomistic 11 31 12 26 11 Fig. 10. structure is defined so that e11 is positive, as in Ref. 20 and e12 is shown in Fig. 10 for a direct comparison with The non-zero direct piezoelectric coefficients are d11 = e11C11−e12C12 e12C11−e11C12 e11 the PBE results from Ref.5 (in which an overall negative 2 2 , d12 = 2 2 , d26 = − , and C11−C12 C11−C12 C66 e31 sign appears related to their choice of X direction, which d31 = (d32 = d31). The magnitudes of d12 and C11+C12 corresponds to the −X direction in Ref. 20 and in the d31 are listed in TableVII, too. Coefficients e31 and d31 present work). The remaining non-zero coefficients are supplement the in-plane coefficients reported in previous 15

2 2 0.0072 data ∆E(ε ,ε )=(67.73 eV)ε ∆E(ε ,ε )=(68.93 eV)ε fit 11 11 11 11 11 11 ∆ ε ε ε 2 0.0048 E( 11, 11)=(69.59 eV) 11

0.0024 SiS SiSe SiTe 0.0000

2 2 2 0.0072 ∆E(ε11,ε11)=(62.04 eV)ε11 ∆E(ε11,ε11)=(56.40 eV)ε11 ∆E(ε11,ε11)=(60.55 eV)ε11

0.0048

0.0024 GeS GeSe GeTe 0.0000

2 2 2 0.0072 ∆E(ε11,ε11)=(53.91 eV)ε11 ∆E(ε11,ε11)=(52.73 eV)ε11 ∆E(ε11,ε11)=(54.48 eV)ε11 ∆ E (eV)

0.0048

0.0024 SnS SnSe SnTe 0.0000

2 2 2 0.0072 ∆E(ε11,ε11)=(46.15 eV)ε11 ∆E(ε11,ε11)=(42.60 eV)ε11 ∆E(ε11,ε11)=(45.50 eV)ε11

0.0048

0.0024 PbS PbSe PbTe 0.0000 -0.01 -0.005 0.00 0.005 0.01 -0.01 -0.005 0.00 0.005 0.01 -0.01 -0.005 0.00 0.005 0.01 (ε ε ) 11, 11

FIG. 12. Quadratic fitting of energy versus biaxial strain. work.5 tric polarization and the piezoelectric response; changes on the out-of-plane electric polarization due to in-plane strain have not been provided before. This study en- VIII. CONCLUSIONS riches the knowledge of the physical properties of two- dimensional group-IV monochalcogenides. The following contributions are contained in the present work: (a) We determined the group symmetry of group-IV monochalcogenide monolayers with a buck- IX. ACKNOWLEDGMENTS led honeycomb structure. (b) We expanded the list of metastable group-IV monochalcogenide monolayers with This work was funded by an Early Career Grant from a honeycomb structure to all twelve compounds on this the U.S. Department of Energy, Office of Basic En- material family: SiS, SiSe, SiTe, GeS, GeSe, GeTe, SnS, ergy Sciences (Award DE−SC0016139). Calculations SnSe, SnTe, PbS, PbSe, PbTe. (c) The electronic band were performed at Cori at NERSC, a U.S. DOE Of- structure and electronic band gap were contributed, too. fice of Science User Facility operated under Contract No. (d) We also contributed the out-of-plane intrinsic elec- DE−AC02−05CH11231. 16

8 2 e12 e31 4 1

0 0 (pC/m) (pC/m)

1 SiS SnS 3 SiSe SnSe ∆ P SiTe SnTe ∆ P -4 GeS PbS -1 GeSe PbSe GeTe PbTe -8 -2 -0.01 0.00 0.01 -0.01 0.00 0.01 ε ε 22 11

FIG. 13. Non-zero piezoelectric responses of group-IV monochalcogenide monolayers with a buckled honeycomb structure.

TABLE VII. Relaxed-ion piezoelectric coeﬃcients e12 and e31 for group-IV monochalcogenides with a (P3m1) buckled honeycomb structure. Direct piezoelectric coeﬃcients d12 and d31 are listed as well. When available, additional data from the literature was added for a direct comparison. ¯ pC pC pm pm Compound Z e12 ( m ) e31 ( m ) d12 ( V ) d31 ( V ) SiS 15 −558 −92 −13.60 −2.24 SiSe 24 −515 −95 −15.95 −2.93 SiTe 33 −425 −90 −14.17 −3.00 GeS 24 −560 −121 −19.67 −4.26 225∗ — 5.65∗ — GeSe 33 −554 −103 −24.67 −4.58 191∗ — 4.88 — GeTe 42 −449 −101 −19.78 −4.43 SnS 33 −574 −135 −29.07 −6.82 164 — 5.28 — SnSe 42 −537 −125 −30.67 −7.12 141 — 4.63 — SnTe 51 −475 −116 −31.56 −7.68 PbS 49 −567 −150 −37.15 −9.79 PbSe 58 −550 −145 −46.72 −12.30 PbTe 67 −490 −126 −37.72 −9.68 ∗: PBE, VASP; Ref.5.

∗ [email protected] 6 Y. Ji, M. Yang, H. Dong, T. Hou, L. Wang, and Y. Li, † [email protected] Nanoscale 9, 8608 (2017). 1 N. Mounet, M. Gibertini, P. Schwaller, D. Campi, 7 D. Gu, X. Tao, H. Chen, W. Zhu, Y. Ouyang, and Q. Peng, A. Merkys, A. Marrazzo, T. Sohier, I. E. Castelli, A. Ce- Nanoscale 11, 2335 (2019). pellotti, G. Pizzi, and N. Marzari, Nat. Nanotech. 13, 246 8 K. Chang, J. Liu, H. Lin, N. Wang, K. Zhao, A. Zhang, (2018). F. Jin, Y. Zhong, X. Hu, W. Duan, Q. Zhang, L. Fu, Q.-K. 2 S. Haastrup, M. Strange, M. Pandey, T. Deilmann, P. S. Xue, X. Chen, and S.-H. Ji, Science 353, 274 (2016). Schmidt, N. F. Hinsche, M. N. Gjerding, D. Torelli, P. M. 9 N. Higashitarumizu, H. Kawamoto, C.-J. Lee, B.-H. Lin, Larsen, A. C. Riis-Jensen, J. Gath, K. W. Jacobsen, J. J. F.-H. Chu, I. Yonemori, T. Nishimura, K. Wakabayashi, Mortensen, T. Olsen, and K. S. Thygesen, 2D Materials W.-H. Chang, and K. Nagashio, Nat. Commun. 11, 2428 5, 042002 (2018). (2020). 3 G. Tritsaris, B. Malone, and E. Kaxiras, J. Appl. Phys. 10 K. Chang, F. K¨uster,B. J. Miller, J.-R. Ji, J.-L. Zhang, 113, 233507 (2013). P. Sessi, S. Barraza-Lopez, and S. S. P. Parkin, Nano Lett. 4 A. K. Singh and R. G. Hennig, Appl. Phys. Lett. 105, (accepted); arXiv:2004.03884 (2020). 042103 (2014). 11 P. Sutter, S. Wimer, and E. Sutter, Nature 570, 354 5 T. Hu and J. Dong, Phys. Chem. Chem. Phys. 18, 32514 (2019). (2016). 17

12 G. A. Tritsaris, B. D. Malone, and E. Kaxiras, J. Appl. (2016). Phys. 113, 233507 (2013). 41 M. Mehboudi, B. M. Fregoso, Y. Yang, W. Zhu, A. van der 13 A. S. Rodin, L. C. Gomes, A. Carvalho, and A. H. Cas- Zande, J. Ferrer, L. Bellaiche, P. Kumar, and S. Barraza- tro Neto, Phys. Rev. B 93, 045431 (2016). Lopez, Phys. Rev. Lett. 117, 246802 (2016). 14 J.-H. Yang, Y. Zhang, W.-J. Yin, X. G. Gong, B. I. Yakob- 42 S. Barraza-Lopez, T. P. Kaloni, S. P. Poudel, and P. Ku- son, and S.-H. Wei, Nano Lett. 16, 1110 (2016). mar, Phys. Rev. B 97, 024110 (2018). 15 M. Qiao, Y. Chen, Y. Wang, and Y. Li, J. Mater. Chem. 43 G. Mills and H. Jónsson, Phys. Rev. Lett. 72, 1124 (1994). A 6, 4119 (2018). 44 G. Henkelman, B. P. Uberuaga, and H. Jónsson, J. Chem. 16 Z. Zhu and D. Tománek, Phys. Rev. Lett. 112, 176802 Phys. 113, 9901 (2000). (2014). 45 D. Liu, X. Lin, and D. Tománek, Nano Lett. 18, 4908 17 P. Rivero, J.-A. Yan, V. M. Garc´ıa-Suárez,J. Ferrer, and (2018). S. Barraza-Lopez, Phys. Rev. B 90, 241408 (2014). 46 P. Hänggi,P. Talkner, and M. Borkovec, Rev. Mod. Phys. 18 S. Cahangirov, M. Topsakal, E. Aktürk,H. S¸ahin, and 62, 251 (1990). S. Ciraci, Phys. Rev. Lett. 102, 236804 (2009). 47 A. Pacheco-Sanjuan, T. B. Bishop, E. E. Farmer, P. Ku- 19 Y. Xu, B. Yan, H.-J. Zhang, J. Wang, G. Xu, P. Tang, mar, and S. Barraza-Lopez, Phys. Rev. B 99, 104108 W. Duan, and S.-C. Zhang, Phys. Rev. Lett. 111, 136804 (2019). (2013). 48 P. B. Littlewood, J. Phys. C: Solid State Phys. 13, 4855 20 K.-A. N. Duerloo, M. T. Ong, and E. J. Reed, J. Phys. (1980). Chem. Lett. 3, 2871 (2012). 49 Ji, S. Cai, T. R. Paudel, H. Sun, C. Zhang, L. Han, Y. Wei, 21 W. Wu, L. Wang, Y. Li, F. Zhang, L. Lin, S. Niu, D. Ch- Y. Zang, M. Gu, Y. Zhang, W. Gao, H. Huyan, W. Guo, enet, X. Zhang, Y. Hao, T. F. Heinz, J. Hone, and Z. L. D. Wu, Z. Gu, E. Y. Tsymbal, P. Wang, Y. Nie, and Wang, Nature 514, 470 (2014). X. Pan, Nature 570, 87 (2019). 22 P. Ares, T. Cea, M. Holwill, Y. B. Wang, R. Roldán, 50 Y. Huang, E. Sutter, J. T. Sadowski, M. Cotlet, O. L. F. Guinea, D. V. Andreeva, L. Fumagalli, K. S. Novoselov, Monti, D. A. Racke, M. R. Neupane, D. Wickramaratne, and C. R. Woods, Adv. Mater. 32, 1905504 (2020). R. K. Lake, B. A. Parkinson, and P. Sutter, ACS Nano 8, 23 J. Q. Toledo-Mar´ın and G. G. Naumis, J. Chem. Phys. 10743 (2014). 146, 094506 (2017). 51 S. Lee, S. Shin, G. Ham, J. Lee, H. Choi, H. Park, and 24 R. M. Pick, M. H. Cohen, and R. M. Martin, Phys. Rev. H. Jeon, AIP Advances 7, 045307 (2017). B 1, 910 (1970). 52 L. Xu, P. Zhang, H. Jiang, X. Wang, F. Chen, Z. Hu, 25 G. Kresse and J. Furthmüller, Phys. Rev. B 54, 11169 Y. Gong, L. Shang, J. Zhang, K. Jiang, and J. Chu, Small (1996). 15, 1904116 (2019). 26 G. Kresse and D. Joubert, Phys. Rev. B 59, 1758 (1999). 53 J. M. Gonzalez and I. I. Oleynik, Phys. Rev. B 94, 125443 27 K. Berland, V. R. Cooper, K. Lee, E. Schröder,T. Thon- (2016). hauser, P. Hyldgaard, and B. I. Lundqvist, Rep. Prog. 54 T. Hahn, ed., International Tables for Crystallography Vol- Phys. 78, 066501 (2015). ume A: Space-group symmetry, 5th ed. (Springer, Dor- 28 J. Klimeˇs,D. R. Bowler, and A. Michaelides, J. Phys.: drecht, 2002). Condens. Matter 22, 022201 (2009). 55 J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. 29 K. Lee, E. D. Murray, L. Kong, B. I. Lundqvist, and D. C. Lett. 77, 3865 (1996). Langreth, Phys. Rev. B 82, 081101 (2010). 56 S. P. Poudel, J. W. Villanova, and S. Barraza-Lopez, Phys. 30 J. c. v. Klimeˇs,D. R. Bowler, and A. Michaelides, Phys. Rev. Materials 3, 124004 (2019). Rev. B 83, 195131 (2011). 57 R. Fei, W. Li, J. Li, and L. Yang, Appl. Phys. Lett. 107, 31 A. Togo and I. Tanaka, Scr. Mater. 108, 1 (2015). 173104 (2015). 32 R. Resta and D. Vanderbilt, “Theory of polarization: A 58 J. W. Villanova, P. Kumar, and S. Barraza-Lopez, Phys. modern approach,” (Springer, Berlin, 2007) Chap. 2, pp. Rev. B 101, 184101 (2020). 31–68, 1st ed. 59 See Supplemental Material at —— for details of calcu- 33 R. Car and M. Parrinello, Phys. Rev. Lett. 55, 2471 (1985). lations leading to the energy path in Figure 4, and for 34 J. M. Soler, E. Artacho, J. D. Gale, A. Garc´ıa,J. Junquera, the electronic properties of the SiS monolayer across that P. Ordejón, and D. Sánchez-Portal, J. Phys.: Condens. path.. Matter 14, 2745 (2002). 60 M. Noor-A-Alam, H. J. Kim, and Y.-H. Shin, Phys. Chem. 35 J. Junquera, O. Paz, D. Sánchez-Portal, and E. Artacho, Chem. Phys. 16, 6575 (2014). Phys. Rev. B 64, 235111 (2001). 61 R. Saito, G. Dresselhaus, and M. S. Dresselhaus, Physical 36 N. Troullier and J. L. Martins, Phys. Rev. B 43, 1993 Properties of Carbon Nanotubes, 1st ed. (World Scientific, (1991). Singapore, 1998). 37 P. Rivero, V. M. Garcia-Suarez, D. Pereñiguez,K. Utt, 62 H. Landolt, R. Börnstein,H. Fischer, O. Madelung, and Y. Yang, L. Bellaiche, K. Park, J. Ferrer, and S. Barraza- G. Deuschle, Landolt-Börnstein: Numerical Data and Lopez, Comp. Mat. Sci. 98, 372 (2015). Functional Relationships in Science and Technology, Nu- 38 P. Hyldgaard, K. Berland, and E. Schröder, Phys. Rev. B merical Data and Functional Relationships in Science and 90, 075148 (2014). Technology Series No. v. 17 (Springer, 1987). 39 G. Román-Pérezand J. M. Soler, Phys. Rev. Lett. 103, 63 A. V. Krukau, O. A. Vydrov, A. F. Izmaylov, and G. E. 096102 (2009). Scuseria, J. Chem. Phys. 125, 224106 (2006). 40 M. Mehboudi, A. M. Dorio, W. Zhu, A. van der Zande, 64 S. Thomas, K. M. Ajith, and M. C. Valsakumar, J. Phys.: H. O. H. Churchill, A. A. Pacheco-Sanjuan, E. O. Harriss, Condens. Matter 28, 295302 (2016). P. Kumar, and S. Barraza-Lopez, Nano Lett. 16, 1704